The reliability of ATLAS-PA is good to excellent, depending on whether a child takes 1, 2, or all 3 subtests. Items on ATLAS-PA are administered until a minimum level of measurement precision is achieved, equivalent to reliability between .75 and .80 on individual subtests. This typically occurs after a child has taken 8 items. If a child completes all three subtests, the reliability of the overall phonological awareness score is above .90.
ATLAS-PA was validated with 938 children exhibiting typical development and 227 children with disabilities related to speech production. Item response modeling indicated that all three subtests formed a unidimensional construct and that, for the children with typical development, the measure was most effective at estimating phonological awareness knowledge for children who are 3 to 6 years of age. There were no ceiling effects observed for the children with disabilities in the present work, so the measure is also appropriate for older children who have speech production difficulties. Furthermore, we examined whether items worked in the same way for the two groups of children, by looking at differential item functioning. Results showed that differential item functioning was minimal, suggesting that ATLAS-PA performs well for both students with and without disabilities. In addition, scores on ATLAS-PA were moderately to strongly related to other measures of phonological awareness.
Read about how ATLAS-PA was validated here.